32 is too many for the likes of kbl, and in order to insert that many
requests into the ring requires us to declare the first few hung --
understandably a slow and unexpected process. Instead, measure the size
of a singe requests and use that to estimate the upper bound on the
chain length we can use for our test, remembering to flush the previous
chain between tests for safety.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: "Yokoyama, Caz" <caz.yokoyama@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321194031.20240-1-chris@chris-wilson.co.uk
If we are already in the desired write domain of a set-domain ioctl,
then there is nothing for us to do and we can quickly return back to
userspace, avoiding any lock contention. By recognising that the
write_domain is always a subset of the read_domains, and excluding the
no-op case of requiring 0 read_domains in the ioctl, we can infer if the
current write_domain matches the target read_domains, there is nothing
for us to do.
Secondary aspect of this is that we undo the arbitrary fetching and
potential flushing of all pages for a set-domain(.write=CPU) call on a
fresh object -- which was introduced simply because we do the get-pages
before taking the struct_mutex.
References: 40e62d5d6b ("drm/i915: Acquire the backing storage outside of struct_mutex in set-domain")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321161908.8007-2-chris@chris-wilson.co.uk
When we return pages to the system, we ensure that they are marked as
being in the CPU domain since any external access is uncontrolled and we
must assume the worst. This means that we need to always flush the pages
on acquisition if we need to use them on the GPU, and from the beginning
have used set-domain. Set-domain is overkill for the purpose as it is a
general synchronisation barrier, but our intent is to only flush the
pages being swapped in. If we move that flush into the pages acquisition
phase, we know then that when we have obj->mm.pages, they are coherent
with the GPU and need only maintain that status without resorting to
heavy handed use of set-domain.
The principle knock-on effect for userspace is through mmap-gtt
pagefaulting. Our uAPI has always implied that the GTT mmap was async
(especially as when any pagefault occurs is unpredicatable to userspace)
and so userspace had to apply explicit domain control itself
(set-domain). However, swapping is transparent to the kernel, and so on
first fault we need to acquire the pages and make them coherent for
access through the GTT. Our use of set-domain here leaks into the uABI
that the first pagefault was synchronous. This is unintentional and
baring a few igt should be unoticed, nevertheless we bump the uABI
version for mmap-gtt to reflect the change in behaviour.
Another implication of the change is that gem_create() is presumed to
create an object that is coherent with the CPU and is in the CPU write
domain, so a set-domain(CPU) following a gem_create() would be a minor
operation that merely checked whether we could allocate all pages for
the object. On applying this change, a set-domain(CPU) causes a clflush
as we acquire the pages. This will have a small impact on mesa as we move
the clflush here on !llc from execbuf time to create, but that should
have minimal performance impact as the same clflush exists but is now
done early and because of the clflush issue, userspace recycles bo and
so should resist allocating fresh objects.
Internally, the presumption that objects are created in the CPU
write-domain and remain so through writes to obj->mm.mapping is more
prevalent than I expected; but easy enough to catch and apply a manual
flush.
For the future, we should push the page flush from the central
set_pages() into the callers so that we can more finely control when it
is applied, but for now doing it one location is easier to validate, at
the cost of sometimes flushing when there is no need.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Antonio Argenziano <antonio.argenziano@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321161908.8007-1-chris@chris-wilson.co.uk
The timeline->name is only used for convenience in pretty printing the
i915_request.fence->ops->get_timeline_name() and it is just as
convenient to pull it from the gem_context directly. The few instances
of its use inside GEM_TRACE() has proven more of a nuisance than
helpful, so not worth saving imo.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321140711.11190-4-chris@chris-wilson.co.uk
Define a mutex for the exclusive use of interacting with the per-file
context-idr, that was previously guarded by struct_mutex. This allows us
to reduce the coverage of struct_mutex, with a view to removing the last
bits coordinating GEM context later. (In the short term, we avoid taking
struct_mutex while using the extended constructor functions, preventing
some nasty recursion.)
v2: s/context_lock/context_idr_lock/
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321140711.11190-2-chris@chris-wilson.co.uk
In later patches, it became apparent that userspace can see a partially
constructed GEM context and begin using it before it was ready, to much
hilarity. Close this window of opportunity by lifting the registration of
the context with userspace (the insertion of the context into the filp's
idr) to the very end of the CONTEXT_CREATE ioctl.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321140711.11190-1-chris@chris-wilson.co.uk
The mock_context() function returns NULL on error, it doesn't return
error pointers.
Fixes: 85fddf0b00 ("drm/i915: Introduce a context barrier callback")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321092451.GK2202@kadam
gcc-4.8 and older dislike the use of __builtin_constant_p() within a
constant expression context, and so we must use the magical
__is_constexpr() instead.
For example, with gcc-4.8.5:
../drivers/gpu/drm/i915/i915_reg.h:167:27: error: first argument to ‘__builtin_choose_expr’ not a constant
../include/linux/build_bug.h:16:45: error: bit-field ‘<anonymous>’ width not an integer constant
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Reported-by: Uma Shankar <uma.shankar@intel.com>
Fixes: baa09e7d2f ("drm/i915: use REG_FIELD_PREP() to define register bitfield values")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190320154021.5244-1-chris@chris-wilson.co.uk
This allows us to ditch i915 in some more places.
v2: use local var in check_vgpu (Paulo)
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190319183543.13679-9-daniele.ceraolospurio@intel.com
This will allow futher simplifications in the uncore handling.
v2: move register access setup under uncore (Chris)
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190319183543.13679-8-daniele.ceraolospurio@intel.com
Move the init, fini, prune, suspend, resume function to work on
intel_uncore instead of dev_priv.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190319183543.13679-5-daniele.ceraolospurio@intel.com
Now that the internal code all works on intel_uncore, flip the
external-facing interface.
v2: fix GVT.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190319183543.13679-4-daniele.ceraolospurio@intel.com
Get/put functions used outside of uncore.c are updated in the next
patch for a nicer split.
v2: use dev_priv where we still have it (Paulo)
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190319183543.13679-3-daniele.ceraolospurio@intel.com
Upper bits are reserved on gen6, so no issue if we write them. Note that
we're already doing this in the non-MT case of IVB, which uses the same
register.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190320122732.14512-1-chris@chris-wilson.co.uk
This patch fixes the PORT_SYNC_MODE_MASTER_SELECT macro
to correctly do the left shifting to set the port sync
master select correctly.
I have tested this fix on ICL.
Fixes: 49edbd4978 ("drm/i915/icl: Define TRANS_DDI_FUNC_CTL DSI registers")
Cc: Madhav Chauhan <madhav.chauhan@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: <stable@vger.kernel.org> # v5.0+
Signed-off-by: Manasi Navare <manasi.d.navare@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190319221847.21311-1-manasi.d.navare@intel.com
Switch to bitmap_zalloc() to show clearly what we are allocating.
Besides that it returns pointer of bitmap type instead of opaque void *.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190304092908.57382-2-andriy.shevchenko@linux.intel.com
skl_update_pipe_wm() is quite pointless now. Just inline it into
skl_compute_wm().
v2: s/skl_build_pipe_wm/skl_update_pipe_wm/ in the commit message (Matt)
Cc: Neel Desai <neel.desai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190312205844.6339-10-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Clean up skl_allocate_pipe_ddb() a bit by moving the 'wm' variable
to tighter scope. We'll also consitify it where appropriate.
Also initialize plane_alloc/uv_plane_alloc when decrlaring them
rather than later.
v2: Update commit message (Matt)
Cc: Neel Desai <neel.desai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190312205844.6339-8-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Currently we disable all the watermarks above the selected max
level for every plane. That would mean that the cursor's watermarks
may also get modified when another plane causes the selected
max watermark level to change. That is not so great as we would
like to keep the cursor as indepenedent as possible to avoid
having to throttle it in resposne to other plane activity.
To avoid that let's keep the watermarks enabled even for levels
above the max selected watermark level, iff the plane has enough
ddb for that particular level. This way the cursor's enabled
watermarks only depend on the cursor itself. This is safe because
the hardware will never choose to use a watermark level unless
all enabled planes have also enabled that level.
Cc: Neel Desai <neel.desai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190312205844.6339-7-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
We use a fixed ddb allocation for the cursor. Now the calculation
actually makes sure we have enough ddb space, but let's double check
anyway.
Cc: Neel Desai <neel.desai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190312205844.6339-6-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Currently we just assume that 32 or 8 blocks of ddb is sufficient
for the cursor. The 32 might be, but the 8 is certainly not. The
minimum we need is at least what level 0 watermarks need, but that
is a bit restrictive, so instead let's calculate what level 7
would need for a 256x256 cursor. We'll use that to determine the
fixed ddb allocation for the cursor. This way the cursor will never
be responsible for missing out on deeper power saving states.
v2: Loop to make sure this works even if some wm levels are
totally disabled (latency==0)
Cc: Neel Desai <neel.desai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com> #v1
Link: https://patchwork.freedesktop.org/patch/msgid/20190319160311.23529-1-ville.syrjala@linux.intel.com
Extract the meat of skl_compute_plane_wm_params() into a lower
level helper that doesn't depend on the plane state. We'll
reuse this for the cursor ddb allocation calculations.
Cc: Neel Desai <neel.desai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190312205844.6339-4-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
skl_compute_plane_wm() doesn't actually need the plane state. While
it would make logically sense to pass it, we shall need to reuse
skl_compute_plane_wm() to compute the minimum ddb allocation for
the cursor before the cursor may be enabled. Thus we can't rely
on the plane state. The alternative would be to duplicate a lot of
the wm calculations for the cursor ddb allocation case, which doens't
appeal to me.
Cc: Neel Desai <neel.desai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190312205844.6339-3-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
If the minimum required ddb space for all the planes equals the
total ddb space available we are allowed to use the relevant
watermark level.
Cc: Neel Desai <neel.desai@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190312205844.6339-2-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
To allow unsetting .is_mobile for the desktop variant
of PNV fix up the cdclk code to select the mobile HPLLVCO register
for both PNV variants.
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190318165633.28924-5-ville.syrjala@linux.intel.com
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
We want to allow the desktop PNV to not have .is_mobile set. To
that end let's add a small helper to determine if the platform
has the ASLE interrupt (or equivalent). Supposdely both PNV
variants have it.
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190318165633.28924-4-ville.syrjala@linux.intel.com
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Add a small helper to determine if we have the panel power
sequencer or not. We'll make PNV an exceptional case so
that we can unset .is_mobile for the desktop variant.
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190318165633.28924-3-ville.syrjala@linux.intel.com
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Make the code self-documenting by introducing i9xx_has_pfit().
Also make PNV an exceptional case so that we can unset
.is_mobile for the desktop variant.
v2: s/gen4/gen>=4/ (Tvrtko)
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190319142329.22881-1-ville.syrjala@linux.intel.com
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
g33/i964g/g45 are the exceptional cases when it comes to
the swizzle detection. Let's reorder the code to handle
them first and let everything else be handled by the
else branch. This allows us to unset .is_mobile for the
desktop PNV variant (which supposedly must follow the
"mobile" path here).
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190318165633.28924-1-ville.syrjala@linux.intel.com
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Exercise acquiring and releasing forcewake around register reads. In
order to read a register behind a GT powerwell, we need to instruct that
powerwell to wake up using a forcewake. When we no longer require the GT
powerwell, we tell the GT to release our forcewake. Inside the
forcewake, the register read should work but outside it should just
return garbage, 0 being the most common garbage. Thus we can detect when
we are inside and outside of the forcewake with just a simple register
read, and so can verify that the GT powerwell is released when we say
so.
v2: Picking the right forcewaked register to return 0 outside of
forcewake is an art.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190320080052.27273-1-chris@chris-wilson.co.uk
Comet Lake is a Intel Processor containing Gen9
Intel HD Graphics. This patch adds the initial set of
PCI IDs. Comet Lake comes off of Coffee Lake - adding
the IDs to Coffee Lake ID list.
More support and features will be in the patches that follow.
v2: Split IDs according to GT. (Rodrigo)
v3: Update IDs.
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190318200133.9666-1-anusha.srivatsa@intel.com
There is probably a issue in DMC firmwares(icl_dmc_ver1_07.bin and
kbl_dmc_ver1_04.bin at least) that causes PSR2 SU to fail after
exiting DC6 if EDP_PSR_TP1_TP3_SEL is kept in PSR_CTL, so for now
lets workaround the issue by cleaning PSR_CTL before enable PSR2.
v2:
- Updated commit description and comment to state that it may be
a DMC firmware issue (Rodrigo)
- No need to RMW, let's write 0 to PSR_CTL(Dhinakaran)
Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190314230113.6571-1-jose.souza@intel.com
Rather than try to maintain some magic relationship between the link
rates and the index into the wrpll params array let's just store
the link rate in the array itself. Much less fragile.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190207173230.22368-13-ville.syrjala@linux.intel.com
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
We already have the code to calculate the WRPLL output clock from
the register values, but for some reason we're only using it for
HDMI and not DP. Throw out the inflexible DP DPLL table lookup and
just call the HDMI code which decodes the actual register values.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190207173230.22368-12-ville.syrjala@linux.intel.com
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>